THE MODELLING OF RADAR SEA CLUTTER
by
SIMON WATTS
A thesis submitted to the
University of Birmingham
for the degree of
DOCTOR OF SCIENCE
The School of Electronic, Electrical & Computer Engineering
University of Birmingham March 2013
University of Birmingham Research Archive
e-theses repository This unpublished thesis/dissertation is copyright of the author and/or third parties. The intellectual property rights of the author or third parties in respect of this work are as defined by The Copyright Designs and Patents Act 1988 or as modified by any successor legislation. Any use made of information contained in this thesis/dissertation must be in accordance with that legislation and must be properly acknowledged. Further distribution or reproduction in any format is prohibited without the permission of the copyright holder.
Abstract
This thesis presents selected research papers and patents written by Simon Watts
between 1985 and 2012 on the topic of the modelling of radar sea clutter. This work
has been based on the development and exploitation of the compound K distribution
model for the amplitude statistics of radar sea clutter. It has covered the
development of the model, through the analysis of recorded radar data, to establish
its validity over a wide range of conditions and for both coherent and non-coherent
radar processing. The work has also developed methods for exploiting the model for
improved performance prediction for radar systems, the analysis and development of
new detection signal processing schemes and the use of these models for the
specification and measurement of radar performance, for the procurement of radar
systems. All the work has been undertaken in an industrial environment, motivated
by the need to develop improved radar systems to meet customer requirements.
Acknowledgements
Much of this work was supported by research funding from the UK MoD, in many
programmes undertaken over the last 25 years. Work was also supported by funding
from the author’s own organisation, Thales UK. The author gratefully acknowledges
this support and also the many other workers who have contributed to this field of
research.
1
Table of contents Table of contents.........................................................................................................1 THE MODELLING OF RADAR SEA CLUTTER..........................................................2 1. Papers presented in the thesis.............................................................................2 2. Biography of Simon Watts....................................................................................5 3. Background to the Research................................................................................5 4. Radar Sea Clutter ................................................................................................6 5. The development and application of sea clutter models by S.Watts ....................8
5.1. Sea clutter amplitude statistics......................................................................9 5.2. The application of sea clutter models to target detection techniques..........10 5.3. Spatial Correlation.......................................................................................15 5.4. Doppler spectra and temporal correlation ...................................................16 5.5. The specification and measurement of radar performance .........................17 5.6. The use of sea clutter models .....................................................................18
6. Conclusions........................................................................................................19 7. References.........................................................................................................19 Selected Research Papers by S.Watts .....................................................................20
2
THE MODELLING OF RADAR SEA CLUTTER
1. Papers presented in the thesis
The papers listed below are the subjects of this thesis. They are referenced by
number in the supporting text and the full papers are appended to the thesis, labelled
Paper 1, Paper 2 etc.
[1] K D Ward, C J Baker and S Watts, “Maritime Surveillance radar part 1: Radar
scattering from the ocean surface” Proc IEE, Vol 137, part F, No.2, April 1990
pp 51-62. (Mountbatten Premium).
[2] S Watts, C J Baker and K D Ward, “Maritime surveillance radar Part 2:
Detection performance prediction in sea clutter” Proc IEE, Vol 137, Par F,
No.2, April 1990 pp 63-72. (Mountbatten Premium).
[3] S.Watts, K.D.Ward and R.J.A.Tough, Modelling the shape parameter of sea
clutter, International Conference Radar 09, Bordeaux, October 2009 (Invited
Paper)
[4] S Watts, “Radar detection prediction in sea clutter using the compound K-
distribution model”, Proc IEE, Vol 132, Part F, No.7, December 1985, pp 613-
620. (J J Thompson Premium).
[5] S Watts, “Radar detection prediction for targets in both K-distributed sea
clutter and thermal noise”. IEEE Trans. AES-23, No.1, January 1987, pp 40-
45.
[6] S Watts, Cell-averaging CFAR gain in spatially correlated K-distributed sea
clutter, IEE Proc. Radar,Sonar,Navig., Vol. 143, No. 5, October 1996, pp 321-
327.
3
[7] S.Watts, Radar performance in K-distributed sea clutter, SET Panel
Symposium on “Low grazing angle clutter: its characterization, measurement
and application”, Laurel, Maryland USA, 25-27 April 2000.
[8] S.Watts, The performance of cell-averaging CFAR systems in sea clutter,
IEEE International Conference, Radar 2000, 7-12 May 2000, pp 398-403.
[9] S. Watts, K.D. Ward and R.J.A. Tough , CFAR loss and gain in K-distributed
sea-clutter and thermal noise, IET International Conference on radar Systems,
Radar 07, October 2007, IET Conference Proceeding 530, Session 7b.
[10] S. Watts, K.D. Ward and R.J.A. Tough, The Physics and Modelling of Discrete
Spikes in Radar Sea Clutter, International Radar Conference Radar 2005,
Washington, May 2005.
[11] S.Watts, UK 2191052 (1981) “Improved cell-averaging CFAR for maritime
surveillance radar”.
[12] S.Watts, UK 2449884, “CFAR control system”.
[13] S.Watts, UK 2297213 (17 Jan 1996);USA 5703592; France 9600418: “Method
and apparatus for estimating the detection range of a radar”.
[14] S Watts and K D Ward, “Spatial correlation in K-distributed Sea Clutter”, Proc
IEE Part F, Vol 134, No.6, October 1987, pp 526-532.
[15] S.Watts, Modeling and simulation of coherent sea clutter, IEEE Trans AES
Vol. 48, No.4, October 2012, pp. 3033-3317
[16] S.Watts, Duration of radar false alarms in band-limited Gaussian noise, IEE
Proc. Radar,Sonar,Navig., Vol. 146, No. 6, Dec 1999, pp 273-277.
4
[17] S.Watts, Specification and measurement of performance for airborne maritime
surveillance radars, International Conference, Radar 99, Brest, 17-21 May,
1999, Session 2.6.
[18] S.Watts, H.D.Griffiths, J.R.Holloway, A.M.Kinghorn, D.G.Money, D.J.Price,
A.M.Whitehead, A.R.Moore, M.A.Wood, D.J.Bannister, The specification and
measurement of radar performance – future research challenges, Journal of
Defence Science, Volume 8, No.2, May 2003, pp 83-91.
[19] S.Watts, H.D.Griffiths, J.R.Holloway, A.M.Kinghorn, D.G.Money, D.J.Price,
A.M.Whitehead, A.R.Moore, M.A.Wood, D.J.Bannister, The specification and
measurement of radar performance, IEE International Conference Radar
2002, 14-17 October 2002, IEE Publication No.490, pp 542-546.
[20] K.D.Ward and S.Watts, Use of sea clutter models in radar design and
development, IET Radar Sonar Navig.,Vol. 4, Iss. 2, April 2010, pp. 146 – 157
(Special Issue on Radar Clutter)
[21] S.Watts, Radar Sea Clutter Modelling – Recent Progress And Future
Challenges, International Conference Radar 08, Adelaide, September 2008
(Invited Paper)
5
2. Biography of Simon Watts
Simon Watts graduated from the University of Oxford in 1971, obtained an MSc from
the University of Birmingham in 1972 and a PhD from the CNAA in 1987. He is
currently deputy Scientific Director and Technical Fellow in Thales UK and is also a
Visiting Professor in the department of Electronic and Electrical Engineering at
University College London. He joined Thales (then EMI Electronics) in 1967 and
since then has worked on a wide range of radar and EW projects, with a particular
research interest in maritime radar and sea clutter. He is author and co-author of
over 50 journal and conference papers, an IET book on sea clutter (with K.D.Ward
and R.J.A.Tough) and several patents. He was chairman of the international radar
conference RADAR-97 in Edinburgh UK. Simon Watts serves on the IEEE AESS
Radar Systems Panel, is an Associate Editor for Radar for the IEEE Transactions
AES and a member of the Editorial Board of IET Radar, Sonar & Navigation. He was
appointed MBE in 1996 for services to the UK defence industry and is a Fellow of the
Royal Academy of Engineering, Fellow of the IET, Fellow of the IMA and Fellow of
the IEEE.
3. Background to the Research
The research described here was mainly undertaken in support of the development
of airborne maritime surveillance radars, originally by EMI Electronics and
subsequently, through the process of industrial acquisitions, by Racal and Thales
UK. Much of the work was undertaken in collaboration with researchers at the UK
government Radar Research Establishment (RRE) and its successors (RSRE, DRA,
DERA, Dstl), either complementing their work or working directly with them. In
6
particular, the research was applied to upgrades to the Searchwater radar, in service
in RAF Nimrod MR2 aircraft from 1980 to 2012, and to the design of the Searchwater
2000 family of radars. The Searchwater 2000AEW is currently in service in the RN
Sea King Mk7 Airborne Surveillance and Control (ASaC) helicopter and the
Searchwater 2000MR was successfully developed and delivered to the RAF for the
Nimrod MRA4 aircraft, which was cancelled in 2012. Although this research was
motivated to support the development of airborne maritime surveillance radars by the
company, it has had application worldwide in the development of all forms of radar
operating in a maritime environment. The emphasis of all the work has been to
develop models and techniques that have practical application to all aspects of radar
design and manufacture. The work spans the period from the early 1980s to the
present day.
4. Radar Sea Clutter
A radar operating in a maritime environment will usually observe signals reflected
from the sea surface, in addition to reflections from targets of interest, which may
vary from large ships to very small targets such as submarine periscopes. The
signals reflected from the sea, known as radar sea clutter, are usually unwanted and
may significantly interfere with the signals from wanted targets.
A prime example of a military application that encounters problems of this kind is
maritime surveillance. Typical examples include the Searchwater radar, which was in
service with the UK RAF Nimrod MR2 aircraft, the AN/APS-137 radar fitted in the US
Navy P3-C aircraft, and the Blue Kestrel radar in the Royal Navy Merlin helicopter.
These radars have many operating modes, but in particular are used for long-range
7
surveillance of surface ships (known as ASuW, anti-surface warfare) and detection of
small surface targets, including submarine masts (ASW, anti-submarine warfare).
For ASuW operation, the radar must detect, track and classify surface ships at
ranges in excess of 100 nautical miles. For ASW operation, the radar must detect
submarine masts just above the surface in high sea states, at ranges of many miles.
In both these modes, the radar operator and the automatic detection processing must
be able to distinguish between returns from wanted targets and those from the sea
surface. Unlike satellite radars, the grazing angle at the sea surface for such radars
is typically less than 10° and often the area of interest extends out to the radar
horizon (i.e. zero grazing angle). Under these conditions, returns from the sea can
often have target-like characteristics and may be very difficult to distinguish from real
targets. In order to aid discrimination between targets and clutter and, in extreme
cases, prevent overload of the radar operator or signal processor, the radar detection
processing must attempt to achieve an acceptable and constant false alarm rate from
the sea clutter.
The design of such signal processing algorithms requires a detailed understanding of
the characteristics of the signals received from both targets and the sea. For this
reason, the characteristics of sea clutter have been, and continue to be, extensively
studied. The aim of these studies is to improve our understanding of these
characteristics, which can vary very widely, dependent on the environmental
conditions, viewing geometry and radar parameters. In particular, the desire is to
build mathematical models that can capture these characteristics and then be applied
to the design and manufacture of better radar systems.
8
Early work on the development of models of sea clutter concentrated simply on
developing models of the mean intensity of the backscatter under different
conditions. Subsequently, attempts were made to models the fluctuations about the
mean intensity. Early models failed to capture the complex statistics of these
fluctuations but a major advance was made by the formulation by K.D. Ward of the
compound K distribution model for the non-Gaussian statistics of sea clutter [A]. This
was the starting point for the research described here.
The compound K distribution model has been specified for use in defining the
required radar performance by defence procurement agencies, incluing those in UK
and USA, and is widely accepted as the definitive model for sea clutter amplitude
statistics. A search on IEEE Xplore for the terms “K distribution” and “sea clutter”
together returns over 4000 results. The model provides the basis for methods for
predicting performance of radars detecting targets in sea clutter that are much more
accurate, over a wide range of conditions, than previous modelling methods. The
model also provides insight into the wider characteristics of sea clutter, supporting
the development of improved radar waveforms and detection signal processing
algorithms.
5. The development and application of sea clutter models by S.Watts
This Section describes the contributions described in the papers presented in this
thesis. The discussion is divided into different aspects of research into the modelling
of radar sea clutter. Section 5.1 covers the modelling of the amplitude statistics of
the backscatter clutter signals received by a radar. Section 5.2 describes the
9
application of these models to the development of improved radar signal processing
algorithms for the detection of small targets against a background of sea clutter
returns. Section 5.3 looks at the specific topic of modelling the spatial correlation of
clutter returns and Section 5.4 covers work on the Doppler spectra and temporal
correlation of sea clutter. An important use of models is to assist in the procurement
of advanced radar systems and Section 5.5 reviews work undertaken on the
specification and measurement of radar performance. Finally, Section 5.6 considers
work that has been done to explain and promote the appropriate use of models of
sea clutter and to promote further work in this area.
5.1. Sea clutter amplitude statistics
The returns from a sea clutter may appear in a radar receiver as noise-like returns.
This noise is generally modelled as a stochastic process, but unlike the thermal noise
also always present at some level in a radar receiver, its statistics are usually non-
Gaussian.
The seminal papers that set out the evidence for the validity of the compound K
distribution model were Ward, Baker, Watts [1] and Watts, Baker, Ward [2], written in
collaboration with K.D. Ward and C.J. Baker, then of RSRE. These papers describe
the characteristics observed in radar data that led to the compound form of the
model. They demonstrated the good fit to the model consistently achieved with a
large amount of real radar data, recorded over a wide range of environmental
conditions and radar parameters. A particularly important result was the model
reported in the paper that related the shape parameter of the K distribution amplitude
PDF to the environmental conditions, the viewing geometry and the radar
10
parameters. This allowed the use of the model for generic radar design applications.
Watts collaborated in all this work and in particular was the lead author for [2], which
showed results for the exploitation of the models in performance prediction (see also
Section 5.2). These two papers were awarded the IET Mountbatten Premium.
Work on the modelling of amplitude statistics has continued over many years, as
further data has been collected. More recently, from the late 1990s to the present
day, developments in radar data recording techniques have allowed some of these
models to be revisited, using data gathered during the development of the
Searchwater 2000 radars. Of particular interest is the paper by Watts, Ward and
Tough [3], which was presented as an invited paper at the Radar 2006 International
Conference. This paper, for which Watts was the lead author, provided greater
insight into the variation of amplitude statistics (i.e. the shape parameter and the
probability density function of the amplitude) with viewing geometry (particularly the
grazing angle), reinforcing the idea of a critical grazing angle, below which clutter
statistics become increasingly non-Gaussian. The work contributing to this part of
the paper was undertaken by Watts. The paper also reported development in the
electromagnetic modelling of the scattering from the sea surface, undertaken by
Ward and Tough.
5.2. The application of sea clutter models to target detection techniques
One of the main purposes for the development of statistical models of sea clutter is
the design of improved detection algorithms. The compound form of the K
distribution model implies two components of the amplitude statistics, that have
11
different spatial and temporal correlations. These must be correctly modelled when
predicting performance and will also guide the design of new detection algorithms.
One of the first papers explaining how to predict detection performance of targets in
K distributed sea clutter for non-coherent radar systems, particularly those employing
pulse-to-pulse frequency agility, was by Watts [4]. This showed how the effects of
pulse-to-pulse analogue integration could be modelled and also described the effects
of binary integration on detection performance. In this paper, the concept of the
“ideal CFAR” (ideal Constant False Alarm Rate) detector was introduced. This
showed how performance would be considerably enhanced if the detector were able
to follow exactly the underlying mean level component of the clutter. This gave an
upper bound to performance and these ideas were developed further in the papers
on the Cell-Averaging Constant False Alarm radar (CA CFAR) technique, discussed
below. The paper [4] was awarded the IET JJ Thomson Premium.
Thermal noise is always present at some level in a radar receiver due to the inherent
noise of the receiver electronics. Although clutter returns may be strongly received in
a radar at short or intermediate ranges, towards the radar horizon the clutter returns
may be very weak or non-existent and radar performance will be limited by the
receiver thermal noise alone. It is very important to be able to model the effects of
the changing clutter-to-noise power ratio as the conditions or viewing geometry
change. Uniquely, the compound K distribution model allows thermal noise to be
accurately incorporated into the modelled sea clutter returns. The method for
incorporating noise and the subsequent calculations of detection performance were
12
first described in Watts [5]. This is an important paper in this field and, together with
[1] and [2], has been widely cited by subsequent researchers.
A well-known technique for setting an adaptive radar detection threshold is known as
the CA CFAR algorithm. This technique predicts the mean level of the radar return of
interest at a particular range, by estimating the mean level of the returns from a
number of adjacent range cells. The CA CFAR was originally developed for use in
thermal noise, but is widely applied for target detection in the presence of sea clutter.
As also discussed in Section 5.3, the returns from sea clutter often exhibit a spatial
correlation in range, related to the presence of sea swell or waves that modulate the
local intensity of the backscattered signal. This spatial correlation can have a very
significant effect on the performance of CA CFAR systems, as the cell-under-test is
no longer statistically independent of the surrounding cells. The analysis of this
performance was initially considered in [2] and then in more detail by Watts [6],
showing how performance is affected and how the changes in performance can be
quantified. In [6], it was shown that in some real clutter conditions, the performance
of the “ideal CFAR” detector, quantified as a “CFAR gain”, could be approached by
appropriate selection of the cell-averager length, dependent on the spatial correlation
properties of the clutter. It was also shown how the selection of cell-averager length
can effect the spatial distribution of false alarms in clutter. A spatially uniform
distribution of alarms is very desirable for efficient radar detection. In some
conditions, the incorrect selection of cell-averager length can result in a localised
bunching of alarms, which can be very deleterious to radar performance.
13
These areas of performance analysis were further developed and explained in Watts
[7] and [8]. In Watts [7], the performance of detectors following pulse-to-pulse
integration was further developed, showing how the extremes of pulse-to-pulse
correlation of the speckle component of the clutter (for example, such as induced by
frequency agile or fixed frequency operation) could be incorporated into performance
predictions calculations. In Watts [8], the performance of the CA CFAR was further
examined, quantifying the CFAR loss or gain, compared to an “ideal” fixed threshold
detector, for different clutter conditions and for different CA CFAR configurations. It
is usual to expect a loss in performance when using a CA CFAR detector, but it was
shown that there will be an optimum length for the cell-averager, which can result in a
CFAR gain in some conditions. This work was further extended in Watts, Ward and
Tough [9]. In this paper, the part written by Watts reported the effects of added
thermal noise to the results reported in [8]. This showed the interesting result that in
some circumstances added thermal noise can reduce the loss of a CA CFAR system
compared to a fixed threshold under the same conditions.
A particularly difficult characteristic of sea clutter that is sometimes observed is the
presence of discrete target-like “spikes”. These spikes may be very large compared
with the surrounding clutter returns and may persist for up to a second or more. This
behaviour is not modelled by the standard compound K distribution, but can be
accommodated in the KA model, first discussed by Middleton [B] and Ward and
Tough [C]. Watts developed this model using recorded data and also showed how
the additional parameters of the model could be estimated from data measurements,
including the effects of added noise. This work was summarised in Watts, Ward and
14
Tough [10]. Again, this work showed how the model could be applied to practical
radar design and discussed the implications of such characteristics to radar detection
performance. In that paper, material on the modelling of electromagnetic scattering
was contributed by Ward and Tough, whilst the development of the KA model was
undertaken by Watts.
One of the aims of this research was always to develop improved detection
algorithms. Where these were likely to be exploited in manufactured equipment,
patent protection was sought.
Watts [11] was patented in GB and describes a novel technique for improving CA
CFAR performance in correlated clutter, by modelling the spatial correlation of the
clutter as an autoregressive process and adjusting the weights on the cell-averager
FIR filters accordingly. This can provide improvements to detection performance,
albeit with added algorithmic complexity.
Watts [12], patented in GB, France and USA, describes improvements to the CA
CFAR detector based on exploiting models of the sea clutter characteristics. The
clutter characteristics are estimated by observing a high false alarm rate at a low
threshold setting. The detector then uses this fit to the model to estimate the
required higher threshold to achieve the desired low false alarm rate, which cannot
be measured directly. This allowed the CA CFAR threshold to adapt to changing
clutter conditions very rapidly, whilst maintaining a good CFAR performance. This
15
circuit has been employed in a number of maritime radar systems sold by the
company.
The algorithms used in a modern maritime radar may be very complex, with a
number of controls and options available to the radar operator (such as CA CFAR
settings, choice of radar antenna polarization, optimum aircraft operating height, and
so on). It may not be immediately obvious to a radar operator that the radar and
viewing geometry have been set up in manner that are optimum, given the prevailing
conditions and targets of interest. The invention described in Watts [13] also exploits
the results of clutter modelling to provide a predicted-detection-range display. This
uses the radar’s own measurements of the conditions, as sensed by its adaptive
circuits, such as the CA CFAR threshold circuit described in [12], sensing the clutter
amplitude statistics, together with automatic gain control (AGC), which senses the
mean level of the clutter, to show the operator on his display what size of target may
be detected at different points of the sea surface under surveillance. As the operator
adjusts the radar, aircraft height and so on, the radar thresholds adapt and the
resulting improvement or degradation of performance is presented to the operator as
changing detection contours on the display. This gives the radar operator confidence
in his settings and allows intelligent testing of different settings to produce the best
performance. Again, the invention has been incorporated in radars manufactured by
Thales.
5.3. Spatial Correlation
Spatial correlation of clutter returns is a specific feature that can impact detection
performance, as discussed above, and so must also be modelled. The spatial
16
correlation observed by a radar is a function of the prevailing conditions (wind waves
and swell), the look direction and the spatial resolution of the radar. An early paper
that described the different spatial correlations of clutter and the impact of this on the
amplitude statistics with different spatial resolutions was described in Watts and
Ward [14]. The work by Watts in this paper showed how spatial correlation could be
modelled, including the effects of changing spatial resolution. The paper also
reported methods for simulating correlated clutter returns in a computer, which was
done by Ward.
5.4. Doppler spectra and temporal correlation
Much of the earlier work discussed above was concerned with the modelling of the
statistics of the envelope or intensity of the clutter returns, which is what has been
processed by most airborne maritime surveillance radars until recently. There is
increasing interest now in exploiting the Doppler shift of maritime targets, to enhance
their detectability against sea clutter. One reason for this is that modern radar
technology can produce the required high quality coherent radar returns relatively
easily compared with the technology available, say, 20 years ago.
Models of Doppler spectra of sea clutter have been much less well developed than
those for amplitude statistics. However, they are no less important for the design and
assessment of modern radar systems. There is a large body of literature that
describes the mathematics of so-called optimum detectors in coherent sea clutter,
based around the Generalised Likelihood Ratio Test (GLRT). Such techniques
assume that the clutter Doppler characteristics can be defined in terms of a
covariance matrix, as a Spherically Invariant Random Process (SIRP). In fact,
17
however, the Doppler characteristics of sea clutter are generally quite non-stationary,
as noted many years ago by Ward, Baker and Watts [1], and cannot be described in
terms of a simple covariance matrix or an SIRP. Recent observations by Watts of the
characteristics of Doppler spectra of sea clutter have led to a new model being
proposed, which captures this non-stationary behaviour. This model allows more
realistic simulation of clutter returns and also provides the basis for improved
detection performance predictions. This work was recently reported by Watts [15].
A rather different analysis of the practical performance of radar detectors was
reported in Watts [16]. This related to the statistics of the duration of alarm crossings
of a threshold in thermal noise. The duration of false alarms is a function of the
frequency response of the radar receiver and the probability of false alarms. The
proper understanding of this characteristic, which was not previously correctly
reported in the radar literature, is important when interpreting the relationship
between the probability of false alarm set by the radar detection threshold, and the
number of alarms seen in a given area of a radar display. As the performance of
radars is scrutinised in ever greater detail, such understanding is essential to correct
assessment of radar performance in practical trials.
5.5. The specification and measurement of radar performance
The procurement of a new advanced radar system by a customer is a complex
process, requiring considerable technical insight by the customer if he is to be able to
acquire a system that properly meets his needs. Such radars are very rarely
supplied “off-the-shelf” with fixed characteristics, but are developed or modified to
meet specific customer requirements. Firstly, the customer must specify what is
18
required and the manufacturer must then translate these requirements into a
mutually acceptable performance specification. Once the radar is developed, its
performance must be demonstrated to the customer so he can accept or reject
delivery of the system against the original specified performance. It is often very
difficult or impossible to prove radar performance by demonstration in practical trials
and often part of any acceptance process must involve modelling. The problem of
specifying and then measuring maritime radar performance was highlighted for the
first time in the radar literature by Watts [17]. This work looked at what constitutes
the performance of radars detecting targets in sea clutter and how such performance
can be quantified and measured. This work was further developed when Watts
proposed and then chaired a working party for the Defence Scientific Advisory
Council (DSAC) of the UK Ministry of Defence. This work involved experts from UK
industry, government research laboratories and academia to provide advice to MoD
on the specification and measurement of advanced radar systems, from the
perspective of the customer procuring such systems. This work was published in
Watts et al. [18] and [19]. The topic continues to be of considerable importance.
5.6. The use of sea clutter models
It is appropriate from time to time to revisit the underlying issues associated with why
models are used and which models are appropriate to use in different circumstances.
The many different uses for models of sea clutter are discussed in Ward and Watts
[20]. This work emphasises the role of models in the specification and measurement
of radar performance. It also highlights the possible misleading results that can be
obtained if the wrong models are used to predict performance in a given situation. A
typical example would be the failure to use a compound formulation for the amplitude
19
statistics, and so not properly modelling the effects the correlations of the different
components on detection performance.
Finally, to encourage the direction of further research, the occasional tutorial on the
state-of-the-art is useful to highlight areas where further research is needed. Such a
summary of the current state of research was given by Watts [21], as an invited
paper at Radar 2008.
6. Conclusions
The body of research presented here comprises papers published between 1985 and
2012. The work is mainly concerned with the modelling of sea clutter, based on the
compound K distribution formulation, and with the exploitation of these models. This
exploitation includes improved performance prediction, the design of improved signal
processing algorithms for target detection and the role of these models in the
specification and measurement of performance for the purposes of radar
procurement.
7. References
A. K. D. Ward, Electronics Letters, 17, 561, 1981
B. D Middleton, New physical-statistical methods and models for clutter and
reverberation: the KA distribution and related probability structures, IEEE J.
Oceanic Eng., Vol 24, No. 3, July 1999, pp 261-284.
C. K D Ward and R J Tough, Radar detection performance in sea clutter with
discrete spikes, International Conference Radar 2002, IEE Publication 490,
15-17 October 2002, pp 253-257
20
Selected Research Papers by S.Watts
Top Related